Automated hyper-parameter optimization for deep learning framework to simulate boundary conditions for wave propagation
Harpreet Kaur, Sergey B. Fomel, Nam H. Pham · 2022
We propose a hyper-parameter optimization workflow for training the deep learning framework to simulate the effect of boundary conditions for wave propagation. Hyper-parameter selection is a crucial step in model building and has a direct impact on the performance of machine learning models. We implement three different hyper-parameter optimization techniques, namely random search, Hyperband, and Bayesian optimization, for the proposed network to simulate boundary conditions and compare the strengths and drawbacks of these techniques. The automated deep learning framework optimizes network training and significantly improves the efficiency of the workflow. The proposed method reduces human effort in the network tuning process, improves the performance of deep learning models by achieving the optimal minima, makes the model more reproducible, and can be extended to different deep learning based applications. Tests on different models verify the effectiveness of the proposed approach.